Dispute De-Escalator - Conflict to Calm

• 

  Instead of advice, the system asks targeted questions to fill critical gaps—forcing clarity and reducing assumptions.

• 

  Gemini 3 enters a reasoning phase—separating emotion from facts, identifying dispute type, and detecting missing information before acting.

• 

  Three de-escalation strategies are generated—from soft communication to formal escalation—with ready-to-use draft templates.

• 

  Users describe disputes in plain language and upload real documents. Gemini Vision analyzes contracts while text models parse the narrative.

• 

  Navigate disputes without court chaos. A Gemini-3 powered system that clarifies facts, detects gaps, and guides safe pre-court resolution.

• 

  A unified intelligence report shows conflict level, resolution potential, and fact clarity—giving users a clear situation overview.

The Inspiration

Small disputes rarely begin as legal battles—they start as a fog of confusion, high emotion, and missing data. We observed that first-time "dispute holders" (tenants, freelancers, or employees) often escalate issues prematurely. They don't necessarily want a fight; they simply lack a clear map of what actually matters in their specific situation.

While most legal-tech tools focus on answering questions users already know how to ask, our insight was different: people often don't know what the right question is in the first place. We built Dispute De-Escalator to help users slow down, structure the facts, and find safe resolution paths before entering a costly legal arena.

What the Project Does

Dispute De-Escalator is a pre-litigation intelligence and de-escalation system, not a legal advice tool. It acts as a neutral "buffer" to turn heated narratives into actionable clarity.

Emotional Intake: Accepts raw narratives via text or voice, allowing users to "vent" while the AI extracts facts.

Multimodal Evidence Review: Analyzes uploaded documents (contracts, screenshots, PDFs) to verify dates, clauses, and parties.

Intelligent Gap Detection: Identifies what information is missing by comparing the story against typical dispute patterns.

Targeted Clarification: Asks a few neutral questions designed to reduce uncertainty rather than provoke confrontation.

Case Intelligence Report: Synthesizes the data into a visual summary of conflict intensity and factual clarity.

Resolution Pathways: Generates three escalating, non-binding options—ranging from professional clarification to formal notice—complete with ready-to-use templates.

The Agentic Workflow

Rather than a standard chatbot response, the application uses a chained reasoning pipeline powered by Gemini 3:

Scenario Understanding: Extracts verifiable facts and detects the emotional "temperature" of the dispute.

Evidence Analysis: Gemini’s vision capabilities parse offer letters and notices to ground the narrative in reality.

The "Thinking" Phase: The system mimics a human consultant, identifying logical gaps in the story.

Strategic Inquiry: Users are prompted for specific details to minimize "he-said-she-said" dynamics.

Synthesis: The system calculates resolution potential and generates a Case Intelligence Report.

Pathways: Provides non-adversarial drafts that help users communicate professionally.

Why Gemini 3 Matters

This project relies on the specific "reasoning" strengths of the Gemini 3 architecture:

Long-Context Reasoning: To maintain a cohesive understanding of a case as more facts emerge.

Multimodal Native Support: To treat a photo of a lease and a text description of a leak as a single, unified data point.

Structured Output: Using JSON to drive the UI logic, ensuring the system behaves as a decision-support agent rather than a conversational toy.

Challenges & Learnings

The Advice Boundary: Designing prompts that provide guidance without crossing into unauthorized legal advice.

Transparency: We learned that showing the AI's "thought process" builds user trust in a way that "magic" answers do not.

Intentional Friction: In de-escalation, speed isn't always the goal. We found that forcing users to pause and answer clarifying questions naturally lowers the emotional stakes.

Disclaimer

This project is a hackathon prototype designed for informational guidance only. It does not provide legal advice, legal judgments, or replace the need for professional legal counsel or court intervention.

Built With

• css
• gemini
• lucide
• mediarecorder
• react
• typescript